[b-hebrew] Origins of the definite article of BHebrew

Herman Meester crazymulgogi at gmail.com
Sun Dec 18 16:50:46 EST 2005

2005/12/18, Peter Kirk <peter at qaya.org>:
> On 17/12/2005 14:50, Herman Meester wrote:
> >...
> >You said that Qur'anic spelling is conservative. Although this, I
> >guess, is right (actually, originally not so much the written, but the
> >spoken, recited text was holy to Muslims), if the ل laam had been
> >there already before some consonants (obviously the case), we first
> >have to know how long this spelling already existed; Arabic wasn't
> >born for the Qur'an only.
> >
> >
> Of course. Basically I am assuming (although I have not checked) that
> the spelling with alif-lam in alll cases was in the oldest surviving MSS
> of the Qur'an, which date back to within a century or two of the time of
> Mohammed if I am not mistaken. I would be very surprised if this does
> not represent a spelling already current in Mohammed's time, but I can't
> be certain of course. Are there in fact any older surviving Arabic MSS,
> of other works? What form of the article is used in them? Of course
> Arabic existed before then, but we have very few if any earlier written
> records.

I'm now trying to find out more about that, because my knowledge is
too limited at the moment. I've been looking for material in the univ.
library, and I'll be consulting Arabists. My feeling is that this
spelling is quite old, but I'll keep you up-to-date ;)

> >A certain spelling may just be very convenient, without representing a
> >historical development in all cases. In Korean script, an alphabet
> >that was designed by scholars, letters are written that have never
> >existed, but they give a hint to the etymology of the preceding
> >syllable. Arabic consonantal spelling represents sounds that are not
> >written (vowels, geminations), why would it not have some signs that
> >do not represent a sound, for a change?
> >
> >
> I don't understand you point here. Are you suggesting that lam was
> originally a sign of gemination? That can hardly be true as it is
> clearly cognate with Hebrew lamed, Greek lambda and our own L, and is
> used elsewhere for an L sound.

No, I think laam is always meant as -L-, be it to be actually
pronounced, or "virtually".
All I do suggest is:

* say we have a type of spelling (i.e. Arabic spelling) where the
pausal form is the way a word is represented, not the context form
(i.e. each word is spelled as if it stands alone). For this we have
thousands of examples all over each Arabic text: - taa marbuuta (in
context, usually pronounced (in classical Arabic grammar) as -t-, but
never written as taa'; - alif al-wasl for stems vii, viii, x (surely
no historical hamz!), and some more.
* this means, we have no attempt at phonetic speling, rather a certain
artificial approach to spelling;
* then we see that there is an "article" that was originally only
* however, because of dissimilation in certain consonants, in quite a
few cases the "article" looks like -l-
* the, in a way very practical, thought comes up (speakers/users of
Arabic not always being interested in comparative linguistics nor in
phonetic spelling, just as it goes with any language): "that laam must
be something real; we may not pronounce it all the time, but let's
write it whenever we think it "is there/could have been there", i.e.
in the definition marker of nouns."
* then we don't really have the Laam as a sign of gemination, we just
have a certain interpretation of the dissimilated laam as the "real"

I'm not saying that the Arabs of antiquity were stupid or that they
didn't know that the laam wasn't really there in the "sun-letters";
the very fact that they so consistently write the laam means exactly
that they are *not* interested in avoiding writing anything
unnecessary; the spelling uses this laam to say: "watch out: definite

Why then is the original article not *l-? Because the very comparative
semitic linguists that think the gemination theory is nonsense, say
that L(aam) doesn't assimilate in Arabic! Slight problem for them...
However, let's take a look at Josef Tropper's article "Die
Herausbildung des bestimmten Artikels im Semitischen" in JSS 46/1
Spring 2001.
First he says (page 7), I translate freely, not manipulatively ;)

"In the Arabic branch [of Semit. lang. HM], the situation of the
definite articles that we find is complex... The article (')al-  [sic,
I wonder why not ('a)l- ? HM] of classical Arabic seems to differ
completely from the epigraphic material of early Northern Arabic
dialects; the latter, like Canaanitic, have as the normal form h-."

Basically, Tropper agrees with me that Arabic "doesn't fit" the
*han-theory ("it seems").
He literally says: "If we want to avoid introducing some "ad hoc"
rule, we are forced to accept that the -l- in the article (')al- is
not original." [!]
However, he solves this "seeming" problem. For Arabic he supposes:

*han ~>  *'an ~> *'anbint ~> 'albint (he calls this, actually, "dissimilation")

explaining all these steps in a rather complex matter -which,
admitted, he already announced- referring to all the things that are
possible, phonetically, in various Arabic dialects.

After I read Tropper's article a while ago, I started to wonder what
the pro *han-theory people think is so unacceptable about the
"gemination-first" hypothesis.
The latter idea is not refuted with arguments by Tropper, he doesn't
say explicitly he disagrees with it.
Tropper doesn't even explain with a single word, if I read him well,
why the consonant "glottal stop" of his *(')al universally
disappeared, not even compensated by lengthening, or what happened to
the -a- vowel (that i.m.o. doesn't really exist either, of course). He
takes this for granted.
On the other hand, his entire theory is based on one half handful of
cases of Ugaritic /hn/ in rather juridical little texts (which in most
languages have their own specific, un-daily style) totally
unconvincing to me, where these /hn/ are not even articles themselves.
Even if we accept that Ugaritic is one of the necessary languages we
have to take into account for assessing all sorts of
(semi)"pan-Semitic" phenomena, it just does *not* fit Arabic in this
case. To just admit this could be a relief to many.

> >As I said before, the dissimilation needs to have been innovated in
> >only one or a few consonants, say ع `ayn or ب baa'; and by analogy it
> >could have been taken over for other consonants too.
> >"Popular etymology" is a notorious cause of historic-linguistic
> >falsification, so people may have thought: "that little sound -l- we
> >hear, it must be something relevant; let's write it down; if we don't
> >hear it, for example "(a)d-dars", but in an analogous form,
> >"(a)l-baab" we do hear it, we'll write it there too! After all it
> >should have been there in the first place." It is very convenient to
> >have the laam there to indicate a gemination. After all Arabic is a
> >stenographic type of script; matres lectionis also developed gradually
> >in the Semitic laguages, for the sake of easier understanding only.
> >
> >
> This process is of course possible, although unlikely simply because as
> you say "Arabic is a stenographic type of script" which avoids writing
> anything unnecessarily. But my point is that written records prove that
> this writing convention must go back to the time of Mohammed, or at the
> very latest shortly afterwards.

Or ages prior to that; the Arabic script was certainly not invented
for the Qur'an; as it is related to Syriac (compare Serto with
Arabic!), is just gradually evolved out of its ancestral alefbet.
Serto/Estrangelo does interesting things with (silent/vowel) "olafs"

However, unlike what you said, Arabic, considering the above (sorry,
repeating myself) does certainly not avoid writing anything
unnecessary; that kind of principle is foreign to it. The principle it
does seem to show is the one mentioned by the various grammars I
consulted and what the medieval grammarians concluded is the rule for
the Qur'an too: spelling is based on the pausal form.
In Qur'anic spelling, too, every "article"-alif is carefully given a
wasla (based on the traditional, well-preserved pronunciation).

If this alif had ever been pronounced as a real hamz in context, I
would be surprised, and with me, a lot of real Arabic grammarians (in
contrast: after a pause, it has always been pronounced, and still is,
as an initial glottal stop, of course: you can't begin a word with a
full vowel, even if you try - that is valid for any language: whenever
you open your mouth to utter a vowel, the start of the stream of air
which forms the vowel is always a consonant!).
And we have to study the classical poetry and its metrical rules in
order to find out if ever a poet took the liberty of turning an alif
al-wasl into an alif al-qat`, for the sake of variation, metre or
rhyme, etcetera. I think this doesn't happen at all, and if I'm right
it means that the thought of inventing such a consonant out of nowhere
didn't come up in the mind of the poet: to the poet, there was only
the sound of the language -Arabic poems, at least in the traditional
compositional techniques, were and are still memorised in their
entirety- the writing down of which was always an option, but the
spelling irrelevant to the poem.

> >For clarity's sake: the alif can represent: 1) the vowel ā; 2) a hamz
> >(glottal stop); 3) a separation at the end of a 3rd person plural
> >perfect (cf. قالوا  qālū) (which is only a straight, vertical line,
> >nevertheless considered to be an alif); 4) an auxiliary vowel (cf.
> >Qur'an 7,155 واختار موسى wa-khtāra Mūsā, with its alifu l-wasli
> >completely unpronounced). All four cases happen all the time.
> >
> >Therefore, I do not understand the tendency to think that in Arabic,
> >everything that's written must be a historical consonant. ...
> >
> But surely most of these alifs, except perhaps for the long A vowel,
> represent historic consonants which have lost their pronunciation.

No, neither the alif al-wasl of article, stem vii, viii, x, the words
ism, ibn, etc., nor the alif at the end of the 3rd person plural
perfect verb (the latter is only a vertical line I think to seperate
words), represent historical consonants.
Unless, as some who defend the *han-theory for the def. article,
suggest that only in the case of Arabic *'al, the hamz on the alif and
its vowel -a- disappeared (while other cases of hamz/alif al-qat` were
simply preserved, or if they were not (a handful cases) always
compensated by vowel lengthening!). Well, of course that suggestion
*could* theoretically be *possible*, but any researcher who is loyal
to his own principles of probability would disregard this as

> >... How many
> >letters in the English word /thought/ are historical? Both -th- and
> >-gh- represent only one historical consonant each, not two. Cf. Dutch
> >/dacht/. As English took an alphabet, the Latin one, that didn't
> >exactly fit, this is quite understandable. Just because for the Arabic
> >language, a distinct-looking script has evolved, it doesn't mean all
> >its consonant-signs represent one consonant each, all the time. This
> >would be quite unique, taking into account spelling systems all over
> >the world. Even Korean's unique, specially invented script is not
> >strictly phonetic: it writes unpronounced letters all the time, only
> >in order to present a word's root in one syllable.
> >
> >
> No, Herman, it would not be unique. All over the world today linguists
> are introducing orthographies for newly written languages which are
> phonetic, or at least phonemic i.e. one letter per meaningfully distinct
> sound. Masoretic Hebrew was a phonetic script. Although we can't be
> sure, we have no reason to think that Phoenician script was not phonemic
> for the consonants. The highly non-phonemic English alphabet is by no
> means typical.

Frankly I always mix up these words "phonemic", "phonetic",
"phonological", etc. (I'll check my dictionary some day) All I can say
is, I believe that not every consonant that was pronounced in a unique
way was written in a unique way too in the various Semitic scripts.
Hebrew has a.o. the `ayin/ghayin and sin/shin issues; ghayin, on the
other hand, is written in Ugaritic; Arabic didn't even distinguish
taa', baa', thaa', (as even their pointing is a medieval invention),
or zaa', raa' and a few others. I believe those alefbets are all
improvised, often rather accurately, but improvised.

> >For the sake of simplicity, hoewever, let's assume, against my
> >conviction, that not the gemination but the ل  l- is the original
> >article. Classical Arabic grammar states quite clearly: the alif of
> >the article (a)l- ال should never be given a hamza! This means that
> >the medieval native Arabic speakers, the grammarians among them, did
> >not think the alif there ever represented a consonant. ...
> >
> >
> I agree with you, there is no glottal stop at the start of the article,
> and of some other words which start with a completely silent alif. This
> is surely the standard understanding of Arabic. And yes, it is another
> exception to the rule that each letter represents an original consonant,
> but I never claimed that this rule was precise in Arabic.

I don't think you did claim that, but it's just a good argument for
the gemination theory.

> >
> >You think it is improbable that Epyptian, rather than "Arabian"
> >Arabic, would have more conservative traits. As I said before: I think
> >there are no linguistic reasons why this would not be possible. A
> >language can be imported or not, this doesn't always matter. In India,
> >English -a- in dance is pronounced in an older pronunciation ([dahns])
> >than in North America: [dehns]. ...
> >
> Don't jump to conclusions about which pronunciation is older. The Indian
> pronunciation is also the standard (RP) British one, but American
> pronunciations quite often represent older British pronunciation,
> preceding some sound changes in the 18th century. The older
> pronunciation is often, as here, preserved in British local accents.
> Yes, this is a case where the migrants and those who have borrowed the
> language have preserved the older pronunciation.
> But my issue here is not the comparison between modern Egyptian Arabic
> and other modern dialects, it is between modern Egyptian Arabic and the
> dialect of the Qur'an, as indicated by the spelling of the article in
> the oldest Qur'an MSS.

I have to take a good look at that. The only Qur'an I have at home is
the traditional text. However, I think the oldest MSS are unvocalised,
so of little use to us: they'll all show the alif, the laam, but no
shadda or wasla, etc. So we have to look for the oldest *vocalised*
mss., not the oldest. However, there are Arabic inscriptions were no
Laam is written in what is obviously the "article", mentioned by
Tropper 2001JSS or Voigt 1998JSS (sorry, I'll check later). To me,
evidence in favour of primary gemination, but turned by the *han-ists
into something that fits their theory as well.

> >...
> >In Egyptian Arabic we have the ج Jiim of several other dialects and
> >Modern Standard Arabic pronounced as Giim, so -G- in "give", not -J-
> >in "jest". In Hebrew the corresponding ג gimel, pronunciation -G- in
> >"give", seems to represent, like Egyptian Arabic ج, the older version
> >(cf. Greek's gamma Γ, also pronounced that way). If so, Egyptian
> >Arabic's double -KK- and -GG- in the article may also preserve the
> >older, not the younger state, in comparison to MSA.
> >
> >
> >
> I take the point. But it seems to me rather likely that  ج Jiim was
> pronounced like English G rather than J at the time of the Qur'an and
> the dispersal of Arabic across the Muslim world, and the pronunciation
> shift to J came later. It is much less likely that Egyptian Arabic
> reversed the pronunciation change.

Right, i.o.w., a possible scenario: the original pronunciation of say,
(a)l-gam` in qur'anic times had been (a)ggam`; in this way it went to
Egypt, and was conserved there all along. In the mean time, before the
g ~> j shift, east of Egypt we see that the dissimilation process has
reached the  giim: (a)ggam` ~> (a)l-gam`. Only after that, there has
been a sound shift  -g- ~> -j-; otherwise (if that sound shift had
preceded the dissimilation) the -j- would just be geminated, just like
-t-, -th-, -d-, -s-, -sh-, -z- etc. that are related phonetically to
-j- but not to -k-.
I'm not sure if I understand this myself ;) but I hope it's correct.
There was plenty of time for this scenario to have taken place; it
looks rather plausible. Provided it fits the mss. too, which I
suspect, but the earliest fully vocalised manuscripts are a few
centuries after the rise of Islam, I think (not the oral tradition of
remembering the Qur'an syllable by syllable, btw.).

כל טוב לך פיתר

More information about the b-hebrew mailing list